A novel way of identifying suspect picture files
Robert Shaw, Anthony S Atkins · 2012
Current procedures for the detection and identification of suspect picture files relies on the use of MD5 hashing and then using the resultant hash value to search a database of known hash values for a match, to identify any suspect file. Even a slight modification to the suspect file causes a different hash value to be created which will then fail to generate a match from the records in the database. A novel approach is proposed by converting the suspect picture file to a sound file, generating a `fingerprint' for this sound wave and using a matching algorithm to identify the file from a suspect file database.